European Union reaches ‘historic’ AI regulation deal

European Union has reached a revolutionary artificial intelligence (AI) policy accord. Before market launch, the newly defined standards expressly impose transparency requirements for models such as ChatGPT and other generic AI systems. This is a huge step forward in the European Union’s approach to the governance and responsibility of AI technology.

Representatives of the European Parliament and Council finalized the agreement and addressed the use of AI by the government for biometric monitoring. They also established guidelines for AI platforms like ChatGPT.

Before entering the market, the agreement places a strong emphasis on transparency requirements, including publishing training material summaries, complying with European Union copyright, and providing technical documentation.

The European Union wants to lead the way in regulating AI by striking a balance between allowing its useful applications and preventing any unintended consequences.

The newly determined agreement, which followed lengthy discussions and negotiations, outlines comprehensive measures for AI models with considerable effect and systemic concerns.

These precautions include the need to evaluate and manage any risks, carry out adversarial testing to improve the resilience of the system, and notify the European Commission of any incidents. 

Balanced Approach to AI Regulation in the European Union

The agreement also requires corporations to guarantee cybersecurity and provide information about how energy-efficient their AI models are. The Parliament regards implementation as vital, as indicated by its commitment to monitoring and encouraging innovative business concepts through sandboxes and establishing effective guidelines for strong AI models.

European Union Commissioner’s Perspective on the Agreement

Following the agreement, Thierry Breton, the European Union Commissioner for Internal Market, described it as “historic,” underlining that the AIAct acts as a springboard for European Union academics and entrepreneurs to take the lead in the global AI race rather than just a set of regulations.

The agreement restricts the use of real-time biometric surveillance by governments to specified instances, such as serious threats in public areas or specific crimes. 

Cognitive behavioral manipulation, social scoring, the extraction of face photos from CCTV or the internet, and biometric systems that predict personal information such as orientation and beliefs are among the prohibited methods. 

Customer Protection and Fines for Violations

Customers can express objections and get explanations for decisions made by AI. Violators receive fines that range from 7.5 million euros ($8.1 million) or 1.5% of revenue to 35 million euros ($37.7 million) or 7% of worldwide turnover. The penalties are well-defined. 

The type of violation and the size of the offending firm determine how severe the sanctions will be. The designated fines intend to offer a tiered reaction commensurate with the seriousness of the infraction and the financial resources of the offending party.

Committees in the European Parliament will vote on the accord at future sessions, and the parliament and council will formally adopt the approved wording before it becomes European Union law.